Rethinking rural water supply policy in the Punjab, Pakistan

WATER RESOURCES RESEARCH, VOL. 29, NO. 7, PAGES 1943-1954, JULY 1993

Rethinking Rural Water Supply Policy in the Punjab, Pakistan

MIR ANJUM ALTAF, 1 DALE WHITTINGTON, 2 HAROON JAMAL, 1 AND V. KERRY SMITH 3

This paper provides an analysis of public policy relating to the rural water supply sector in the Punjab, Pakistan. Based on household survey data, it shows that rural water policies have not kept pace with the rapid economic development in this region and that in the absence of adequate public investment households find private sector alternatives to meet their water needs, often at high economic and environmental cost. Using the contingent valuation method for benefit estimation, it is also shown that household willingness to pay for reliable improved services is much higher than assumed. In fact, full cost recovery is quite feasible in many areas of the Punjab. It is recommended that rural water sector strategy be changed from a centralized, supply-oriented focus to a decentral- ized, demand-oriented policy.

INTRODUCTION

At first glance, a visitor walking around the large agricul- tural villages in the Sheikhupura and Faisalabad districts of the Punjab in Pakistan may think life there was pretty much the same as it was 20 years ago. Women can still be seen in the fields threshing wheat by hand. Buffalo dung is still the primary source of fuel for household cooking, and drying patties cling to the walls of both mud and red brick houses. Horse-drawn carts and rickshaws still fill the streets.

However, many changes soon become obvious. Almost all large villages in these two districts now have electricity, and television antennas can be seen on many roofs. Mechanized wheat threshers and tractors are also increasingly at work in the fields. The community well is no longer used and is likely to be filled with trash. Plastic, fiberglass, and cement water tanks of various shapes and sizes are perched on the rooftops of many houses. Also, in the large villages in the Faisalabad district, perhaps the most surprising thing is the noise, the sound of power looms coming from inside single- story red brick buildings. These small textile enterprises often look like other houses on the street, but inside are small factories with five, ten, or more power looms weaving "gray cloth" (unbleached muslin) for export.

Large villages in the Faisalabad district, and to a some- what lesser extent in the Sheikhupura district, are being rapidly integrated into an export-driven regional economy. More than 50% of households are now primarily engaged in nonagricultural activities. Migrants from other parXs of Pa- kistan are moving to these districts in search of work in textile manufacturing and often settle in large villages where !and and housing prices are cheaper than in urban centers. People from the villages often shop in the nearby cities, and entrepreneurs based in the large urban areas such as Faisal- abad subcontract work to enterprises in the villages. As a

1Applied Economics Research Centre, University of Karachi, Pakistan.

2Departments of Environmental Sciences and Engineering, and City and Regional Planning, University of North Carolina at Chapel Hill.

3Resource and Environmental Economics Program, North Caro- lina State University, Raleigh, and Resources for the Future, Washington, D.C.

Copyright 1993 by the American Geophysical Union.

Paper number 92WR02848. 0043-1397/93/92WR-02848505.00

result, village economies are becoming much more closely linked to urban centers.

The emergence of this Sheikhupura-Faisalabad growth corridor as a major area of economic activity in Pakistan is widely recognized. However, the resulting transformation of life in the villages in these districts (and the extent to which village-level economic activities have contributed to urban and regional growth) has not been as well understood, nor has the importance of these developments for infrastructure investment policy been recognized.

The purpose of this paper is to examine what rapid economic development of large villages in the Sheikhupura and Faisalabad districts of the Punjab implies for water supply policy. Although the pace of economic development in these districts has not been matched throughout the Punjab, man./ other districts have also experienced rapid growth, fueled in large part by the gains in agricultural productivity made possible by the Green Revolution and by remittances from Pakistani workers in the Middle East. Thus

although the large villages in the Sheikhupura and Faisala- bad districts are not typical of all villages, they are at the cutting edge of the economic transformation of the irrigated areas of the Punjab and are likely to be indicative of future developments. The situation in large villages in the Sheikhu- pura and Faisalabad districts thus has special policy signifi- cance because what is now happening in these villages will likely occur in many villages throughout the irrigated parts of the Punjab in the near future.

To study the existing water supply situation in these districts, and to assess households' willingness to pay (WTP) for improved water services, in-depth interviews were con- ducted in 1988 with 756 households in 11 villages: 5 in the Sheikhupura and 6 in the Faisalabad district. In each district some of the villages already had piped water distribution systems installed, and some did not. Table 1 shows the number of interviews completed in each of the study vil- lages, as well as 1981 population figures and selected socio- economic and demographic characteristics of the sample households. (For details of the field procedures see Appen- dix.)

Our survey activities had multiple objectives. First, we sought information that would improve our understanding of the demand for improved water services in these areas. Second, we examined situations where people had the opportunity to choose to connect to public water systems and developed empirical models to describe their decisions. Finally, we conducted surveys in areas where people had

1943

1944 ALrAF ET AL.: RUK•L WATER SUPPLY POLICY IN PUNJAB, PAKISTAN

and did not have access to public systems in order to evaluate the factors that influenced their willingness to pay for different types of public water systems. In this paper, we summarize the insights derived from these analyses and discuss their implications for water policy.

Both the actual connection models and the contingent valuation survey results indicate that households in many villages can afford and are willing to pay more than current tariffs for a connection to a piped water system. However, they will not decide to connect unless the system reliability is dramatically improved.

CURRENT WATER SUPPLY AND SANITATION SITUATIONS IN THE STUDY VILLAGES

Since 1970 a transformation has occurred in village water supply arrangements in the irrigated areas of the Punjab. Historically, most households relied on community wells, shallow ponds, or irrigation canals for all their domestic water needs, and the task of fetching water from sources outside the home was a standard part of women's and children's daily routine. Today community wells have been largely replaced by private .handpumps located inside a household's dwelling or courtyard. In the study villages in both the Sheikhupura and the Faisalabad districts almost all of the households interviewed had their own handpumps that were purchased from, installed by, and maintained by pri- vate-sector vendors.

The villages in the Sheikhupura district are located in what is known as the "sweet water" zone of the Punjab, where groundwater can be used for drinking and other purposes. Groundwater there is found at shallow depths. Because most households have their own well and private handpump, there is little need for these households to collect water from any source outside the home.

The villages in the Faisalabad district are located in the "brackish water" zone. Here the water supply situation is more complicated. 'Water from private handpumps is not usually used for drinking or cooking because it is too salty. However, it can be used for other household purposes such as washing and bathing. Private handpumps are still a valuable convenience to households (as evidenced by the fact that almost every household has one). Here women and children usually collect drinking and cooking water from outside the house. They often walk to handpumps located outside the village, where sweet water can be drawn from alongside irrigation canals, or water may be obtained di- rectly from irrigation canals. Some households purchase water from vendors who haul sweet water from canals or

handpumps and deliver it directly to the house. The price of a 20-L container of water delivered to a house is Rs. 1 (U.S. $1 = Rs. 21 in 1990; Rs. 1 = about U.S. $0.05). (See Whittington et al. [1989] and Whittington et al. [1991] for discussion of water vending in other developing countries.)

The shift from community wells to individually owned handpumps is now virtually complete in the Sheikhupura and Faisalabad districts, and a second transition is now underway. Households in both the sweet water and brackish water zones are now demanding better water services than can be provided with a private handpump. They are install- ing electric pumps on their own wells, pumping water into overhead tanks, and then using that water for a variety of conveniences such as indoor plumbing, toilets, and showers.

ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN 1945

TABLE 2. Relationship Between Choice of Service Level, Economic Status, and Education Level of Household

Average Education Level of Most Educated

Sample Households Average Value Average Household Selecting Service of House, Expenditure, Member of Woman in

Service Level Option, % Thousands Rs Rs/Month Household Household

Villages with piped water supply: handpump only domestic connection only handpump and connection handpump and electric pump handpump, electric pump, and connection

Villages without piped water supply: handpump only handpump and electric pump

Villages with piped water supply: handpump only domestic connection only handpump and connection handpump and electric pump handpump, electric pump, and connection

Villages without piped water supply: handpump only handpump and electric pump

Sweet Water Zone

41 49 NA 6 2 1 60 NA 13 2

47 96 NA 8 4 4 138 NA 12 7 7 209 NA 12 7

100

70 66 NA 8 3 30 132 NA 11 5

100

Brackish Water Zone

21 62 1307 5 2 6 67 1308 9 2

40 112 1562 7 3 4 115 1787 8 3

29 145 1822 10 5 100

38 108 1363 8 4 62 181 1928 10 6

100

These households are making investments which replicate the services that would be available from a public piped water distribution system.

The implications of these developments are different for the sweet and brackish water zones. In the sweet water zone, electric pumps enable households to enjoy the amen- ities associated with indoor plumbing, but handpumps al- ready provide an almost equally convenient source of water for drinking and cooking. In the brackish water zone electric pumps and deeper wells permit water to be raised from lower depths, which in many places reduces the salinity of the water withdrawn. In these cases an electric pump not only enables a household to install indoor plumbing, but may also increase the quality of the water provided, thus eliminating the need to fetch drinking and cooking water from outside the house.

Description of Household Water Supply Systems

Table 2 describes how the sample households had ar- ranged various combinations of water service options.

Sweet water zone. In villages without a piped water system 70% of the households had only a handpump; 30% had a handpump and an electric pump. In villages with piped water systems 45% of the households were not connected to the system (41% had a handpump only, and 4% had a handpump and electric pump). Another 47% had both a handpump and a private connection, and 7% had all three options: a handpump, an electric pump, and a connection. 0nly 1% of the households had only a private connection. Of those households living in villages with a piped supply, only 11% had an electric pump. However, in villages without a piped supply, 30% of the households had an electric pump.

Households are less likely to have electric pumps in

villages where a piped water supply is available. This sug- gests that they may consider these service improvements to be substitutes. However, there are limits to this substitution and neither replaces the need for a handpump. Both a private water connection and an electric pump are typically unreli- able (this is true in both the sweet water and brackish zones). The piped water systems may be inoperative for days at a time; when they are operational, they typically run for just 3 or 4 hours/day. The electric pumps are usually more reliable than this, but they still require electricity, which can be unreliable. Handpumps are needed as a standby water source when the power is out.

Brackish water zone. In the brackish water zone, in villages without piped water systems only 38% of the house- holds had only a handpump; 62% had installed an electric pump, more than twice the percentage in the sweet water zone. In villages with piped water systems, only 25% were not connected (21% had a handpump only, and 4% had a handpump and electric pump); 40% had both a handpump and a connection. Most surprising, in villages with piped water systems in the brackish water zone, 29% of the sample households had all three options: a handpump, an electric pump, and a private connection.

Socioeconomic factors. As expected, sample house- holds with multiple water service options are better educated and have greater wealth as measured by the construction value of their houses (i.e., replacement value) than those with only a handpump. For example, in villages in the sweet water zone with piped water systems, for households with only a handpump, the most educated member of the house- hold has, on average, 6 years of schooling. In contrast, for households having a handpump, an electric pump, and a private connection, the level was 12 years. The average

1946 ALTAF ET AL.: RURnL WATER SUPeLY POLICY IN PUNJ^B, PAKISTAN

value of the house for households that have only a hand- pump is Rs. 49,000; for households with a handpump, an electric pump, and a private connection it is Rs. 209,000.

New Developments in Wastewater Disposal Not only are households in our study villages replicating

the services available from a public water system by install- ing electric pumps and overhead storage tanks, but over the last few years households have been rapidly installing flush toilets. All the study villages lack water-borne sewerage systems. Conventional practice has been that people defe- cate in the fields and open areas around the village. As villages grow, this practice becomes less convenient, and many households in the study villages use "latrines," an isolated, enclosed area in each house or compound, and pay private cleaners Rs. 10-20/month to remove the excrement.

The flush toilets that are now being installed are connected to masonry "septic" tanks that are typically built outside the house, in the street. There is no proper drain field associated with these systems. The nightsoil collects in the tank and the liquid effluent spills out into the open drain in the street. Because the installation of these systems is so recent, there is little reliable knowledge of how often such tanks need to be emptied. Most systems with a flush toilet and septic tank cost about Rs. 3000-4000.

The installation of a flush system does not require that a household have an electric pump and indoor plumbing, because water from the handpump can easily be used to flush the toilet. However, typically a household would purchase a toilet and septic tank system after the installation of an electric pump and indoor plumbing. The village elite are typically the first in a community to install flush systems with septic tanks. These are something of a status symbol imply- ing that as incomes increase the general population will follow their lead. Exact estimates are not available on the

number of households in the study villages that have actually installed flush toilets with septic tanks, but the proportion is probably about 10-20% depending on the village.

POLICIES AND INSTITUTIONS GOVERNING

PUBLIC WATER SUPPLY

Rural Water Supply Policy in Pakistan

As in many developing countfids, rural water supply policy in Pakistan is based on the assumption that house- holds in rural areas cannot afford to pay the full costs of improved water systems. As a consequence, it is argued that central government subsidies are required in order to in- crease "coverage." Pakistan's Sixth Five-Year Develop- ment Plan (1983-!987) identified rural water supply as a "neglected sector" and announced an objective of doubling coverage from 22% of the rural population to 44% by the end of 1988. A tentative coverage target of 75% has been set for the end of the Seventh Five-Year Plan in 1993 (the coverage in 1986 was reported by Pasha and McGarry [1989] to be 35%).

Rural water supply investments are justified as poverty- alleviation programs, not as infrastructure necessary to support economic development. Because of the presumption that all capital costs must be met by the central government, the federal government sets criteria to determine how the available federal budget resources should be allocated

among communities and regions of the country. These criteria have generally been based on policymakers' percep. tions of communities' need for improved water supplies with little direct information on what services people themselves want and woul.d be willing to pay for.

The brackish water areas of the Punjab are considered a high priority for improved water services by central planners because of the poor quality of water supplied by private handpumps. The current policy is that villages in brackish areas with populations over 5000 should be provided with a piped distribution system with house connections. If ground- water quality is acceptable, such systems may be based on tubewells. Otherwise, the water source is an irrigation canal, which requires treatment and is more expensive. It is not considered cost-effective to provide villages with popula. tions less than 5000 with house connections. In these cases government policy is to construct a system with public taps located throughout the village. Public water systems are not being constructed in the sweet water areas of the Punjab, on the assumption that private handpumps already meet basic human needs.

For systems with individual private connections house- holds are expected to pay a monthly tariff of between Rs. 10 and Rs. 12. This tariff is simply a flat monthly charge independent of the volume of water used. In addition to the monthly tariff each household is expected to pay a one-time connection fee of about Rs. 80 and must bear the costs of

connecting the house to the distribution system (which would vary with the distance of the household to the water line, but would typically cost about Rs. 500).

Collection of the tariff for private connections is uneven. In some areas most households that are connected pay; in other areas compliance is low. Data on compliance rates in the Punjab are not routinely available. Probably on the order of 60% of the connected households pay the monthly tariff. Households that do not pay are seldom disconnected from the system.

For systems th•tt have only public taps, households are supposed to pay Rs. 5/month. However, this is rarely collected, in part because people generally want house connections and are disappointed when they are only pro- vided with public taps.

Existing Institutional Responsibilities in the Rural Water Sector

The implementation of these policies involves federal, provincial, and, in principle, district involvement. At the federal level the Ministry of Planning and Development (Planning and Development Division, Physical Planning and Housing Section) has the primary responsibility for incorpo- rating rural water supply objectives into overall development plans. The Ministry of Housing and Works (Environment and Urban Affairs Division) has the major responsibility for providing technical oversight of rural water supply pro- grams. At the provincial level the Public Health Engineering Department (PHED) is the agency responsible for the con- struction and initial operation of most rural water systems. Local government authorities (district and union councilsl are supposed to be responsible for the operation and main- tenance of established public water supply systems. The district or union councils are supposed to take over respon- sibility for running their own piped water systems after a

ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN 1947

TABLE 3. Cost to an Average Household of Different Service Options {,Brackish Water Zone)

Service Option

Total Monthly Monthly Total Capital, Capital, O and Monthly,

Rs. Rs. M, Rs. Rs.

(1) handpump a 1000 13 5 18 (2) domestic connection •' 600 c 6 12 d 18 (3) electric pump e 1500 20 20 40 (1 + 2) handpump and domestic connection 1600 19 17 36 (1 + 3) handpump and electric pump 2500 33 25 58 (1 + 2 + 3) handpump, electric pump, and 3100 39 37 76

domestic connection

aAssumes an economic life of 10 years, 10% real interest. Includes cost of shallow well. t'Twenty yec. i's, 10% interest. CConnection fee: Rs. 100; connection costs: Rs. 500. aMonthly tariff paid by household for an unmetered connection. eTen years, 10% interest.

2-year "demonstration period," but this delegation of re- sponsibility rarely occurs. Instead, the PHED usually con- tinues to operate public water systems (and incur the oper- ation and maintenance expenses).

HOUSEHOLD FINANCIAL COSTS OF ALTERNATIVE WATER SUPPLY OPTIONS

Table 3 presents the financial costs to an individual household of a set of different water supply options intended to be representative of the typical situation. The total capital costs of a private handpump and the construction of a shallow well are about Rs. 1000. Maintenance expenses probably average about Rs. 5/month. Assuming an economic life of 10 years and a real interest rate of 10%, the total monthly cost to a household of owning a private handpump is about Rs. 18. Our analysis of the actual connection decisions of households in the two zones suggest implicit rates of time preference ranging from a high of 44% to about 6.5% [see Altaf et al., 1992a]. While there are a number of important qualifying assumptions underlying those esti- mates, they are broadly consistent with the presence of high real interest rates in rural areas of developing countries. For these reasons we selected a 10% rate that is high by the standards of developed economies, but relatively low based on both our implicit estimates and other evidence from developing countries [Fass, 1988].

With the existing tariff and connection fee, the total monthly cost to a household of a private connection to a piped distribution system is about the same as for a hand- pump. The up-front costs (connection fee plus costs of running a water line from the house to the distribution line) are about Rs. 600. The monthly tariff (in the brackish water zone) is Rs. 12. Assuming an economic life of 20 years and a real interest rate of 10%, the total monthly cost of a private water connection is about Rs. 18.

The cost to a household of installing an electric pump on its private well is about Rs. 1500 (more than twice the cost of a private connection); the monthly operation and mainte- nance (excluding electricity) is about Rs. 20. Assuming an economic life of 10 years and a real interest rate of 10%, the to,ta! monthly cost to the household of an electric pump is about Rs. 40. This does not include the cost of an overhead tank (about Rs. 500) or of indoor plumbing (about Rs. 500).

The monthly cost of water and related services to the household obviously increases significantly when the asso-

ciated conveniences are considered as a composite commod- ity. The monthly costs for both a handpump and a connec- tion to a piped distribution system would be Rs. 36; for a handpump and an electric pump the monthly cost would be Rs. 58. Households with all three service options, a hand- pump, an electric pump, and a private connection, pay about Rs. 76/month for water. It is important to acknowledge that most households' willingness to pay for a reliable system likely lies within the bounds defined by these extremes. This conclusion follows because there are a mix of other re-

sponses to periodic disruptions in either the electricity or a public water system that would assure the availability of water. One of these is simply increased storage capacity. Ideally, households would seek the least cost combination. This simple enumeration of methods for acquiring water represents the most direct and perhaps most expensive method for assembling a set of services that would offer a reliable water supply.

How MUCH ARE HOUSEHOLDS WILLING TO PAY

FOR PIPED WATER SERVICES?

Respondents in both the sweet water and brackish water zones were asked a series of structured questions designed to determine how much they would be willing to pay (WTP) for improved water services [Altar et al., 1992b]. House- holds in villages without piped water systems were asked how much they would be willing to pay per month for a private connection (1) to a water system with standard reliability and (2) to a system with improved reliability. Households in villages with piped water systems were asked how much they would pay for a connection to a system with improved reliability. The description of "improved reliabil- ity" provided to the respondent was somewhat different in the two zones. In the sweet water zone the improvement was described as an additional 4 hours of water from the existing system. In the brackish water zone the improved system was described as providing continuous water supply with ade- quate pressure and reliability. (The surveys in the two zones were held 3 months apart. Based on the experience in the sweet water zone the question was modified for the brackish water zone.)

Table 4 reports the responses from households in villages without a piped water supply for a system with standard reliability. Not surprisingly, respondents in the sweet water zone were willing to pay much less than those in the brackish

1948 ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN

TABLE 4. Households' Willingness to Pay (WTP) for Connection to Piped Water System, by Socioeconomic Characteristics (Villages Without Piped Water Supply)

Socioeconomic Characteristics

Sweetwater Zone Brackish Water Zone

Mean Mean

Percentage WTP Bid, Percentage WTP Bid, of Sample Rs/Month of Sample Rs/Month

Years of education of most educated member of household: 0-8 44 15 38 36 9-12 41 21 41 40 >12 15 33 21 47

Construction value of house (Rs.) 0-49,000 38 14 9 33 50,000-99,000 40 20 22 36 100,000-149,000 I0 21 19 38 ->150,000 12 35 50 44

Overall Mean 21 40

WTP bids are for a piped system with standard (i.e., existing) reliability.

water zone (a mean of Rs. 21 versus Rs. 40). In both the sweet water and brackish water zones, households with more educated members gave higher WTP bids on the average than those with less educated members. Also, respondents who lived in more expensive, better quality houses bid higher on the average than those living in less expensive houses. Table 5 reports two models of the deter- minants of households' willingness-to-pay bids, estimated

TABLE 5. Willingness-to-Pay Models Using Maximum Likelihood Estimates for Interval Data

Brackish Water Sweet Water Zone Zone

Ghazi Minara Santpura Bhaddroo Minara Gatwala

Intercept 18.601 31.561 (1.971) (5.637)

Wealth 0.607 0.341 (2.260) (3.001)

Education 0.919 0.558 (2.122) (1.948)

Family Size -0.174 0.109 (-0.419) (0.420)

Proportion adult women -3.965 -10.455 (-0.433) (-6.306)

Age -0. I 15 -0.174 (-0.902) (- 1.991)

External experience - 1.573 0.257 (-0.305) (0.104)

Zero connection charge -2.892 (-0.833)

Attitude toward free 1.119 -0.536 public provision (0.290) (-0.237)

n 111 184

Family size, number of individuals in household; proportion adult women, number of women > 15 years of age relative to family size; age, age of household head; education, number of years of education of most educated member of household; external experience, qual- itative variable = 1 if any male member of household lived outside the village for periods exceeding 6 months, zero otherwise (intended to reflect the extent to which the household is generally informed about water services or urban areas); attitude toward free public provision, qualitative variable = 1 if respondent feels water ought to be provided free by the state; zero otherwise; wealth, replacement price of house; and zero connection charge, willingness-to-pay bid if one-time connection costs were zero. Numbers in parentheses are ratios of coefficients to estimated asymptotic standard errors.

using the responses from villages with neither experience with nor prospects for public systems. The first column reports estimates for respondents from the sweet water zone and the second the brackish zone. The sample pools re- sponses across the two questions (for systems with standard and improved reliability) asked respondents in each area. Each model is estimated using Stewart's [1983] maximum likelihood estimator for interval data. By treating the an- swers to the willingness-to-pay questions as establishing an interval for their true responses we also adjust for the low and high starting points built into the design.

The wealth (i.e., housing cost) and education measures are significant determinants of responses in both areas. None of the other variables are significant influences on the sweet water zone responses. However, for the brackish zone, the proportion of women in the household presumably available to fetch drinking water significantly reduces the stated bids.

In villages that do have piped water systems, some sample households had a private connection and some did not. In a separate analysis [see Altaf et al., 1992a] we investigated whether an economic model could explain these connection decisions. We found that both tadifs and connection fees were significant negative influences on the likelihood of connecting to the public system in both zones. Moreover, we used these models to estimate the maximum tariffs the respondents answering our survey questions should have been willing to pay (i.e., the samples of respondents used for the WTP models in Table 5). The results indicate strong consistency in the two sets of responses [see Smith et al., 1991]. Thus the contingent valuation responses are influ- enced by the variables that economic theory implies should explain their variation across households, and they are consistent with the predicted bids that a model based on actual behavior indicates these households should have made.

Given these plausibility checks, we can consider how bids from households in both types of villages respond to char- acteristics of the water systems offered. In the sweet water zone about 50% of the households in both groups bid more than the existing tariff of Rs. 10 for a connection to a system with improved reliability. Of these households, those al- ready connected offered a mean WTP bid of Rs. 20/month; those not already connected on average bid Rs. 19.

ALTAF ET AL.: RUaAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN 1949

TABLE 6. Comparative Estimated Costs of Existing Private-Sector Services and Projected Public Connections in a Village Without a Piped Water System

Service Option

Typical Village With 5,000 Population (Brackish Water Zone)

Total Monthly Monthly Total Capital, Capital, O and Monthly,

Rs. Rs. M, Rs. Rs.

Piped water system (100% of households connected): Cost to PHED Cost to households Total

Actual current water expenditures Summation of households' WTP bids for

standard system Summation of households' WTP bids for

improved system Estimated revenue from a standard system Estimated revenue from an improved

system Cost of piped water system to PHED if 80%

of households are connected

1,500,000 a 15,000/' 3,800 c 18,800 281,000 a 2,800 2,800

21,600 1,084,000 e 14,100 t' 9,800 23,900

22,500

31,500

36,000 400 g 11,300 11,700 36,000 400 15,800 16,200

1,200,000 12,000 3,000 15,000

Population is 562 houses at 8.9 inhabitants per household. Actual current water expenditures assumes 62% of households have handpump and electric pump, and 38% have only a handpump. Estimated revenue from a standard system based on a tariff of Rs. 25/month, 80% of households connected, and Rs. 80 connection fee. Estimated revenue from a improved system based on a tariff of Rs. 35/month, 80% of households connected, and Rs. 80 connection fee.

aBased on tubewell at Rs. 300/capita capital costs. bAssumes an economic life of 20 years, 10% real interest rate. CAssumes annual operations and maintenance costs equal to 3% of total capital costs (based on cost

data from PHED). aRs. 500 connection costs per household. eCost of electric pump = Rs. 1500; cost of handpump = Rs. 1000. fAssumes an economic life of 10 years, 10% real interest rate. gComputed over 20 years at 10% real interest rate.

In the brackish water zone households in villages with and without piped water systems were also asked about their willingness to pay for a more reliable system. The mean WTP bid for households in villages without piped water systems was Rs. 56 (compared to Rs. 40 for a standard system). Households in villages with piped water systems only bid a mean of Rs. 33. This large difference may be due to the fact that households that already had private connec- tions were dissatisfied with the performance of the existing water system and may have regarded the prospect of a more reliable system skeptically.

THE ECONOMICS OF VILLAGE-LEVEL WATER SUPPLY OPTIONS

Households' efforts to provide themselves with improved water supplies by purchasing services from the private sector entail substantial expenditures in the aggregate. Table 6 presents an estimate of the amount being spent on private provision of water services by households in a hypothetical village in the brackish water zone with a population of 5000 (562 households, assuming 8.9 people/household) without a piped water system. Assuming in accordance with the sur- vey findings that 62% of the households in the village have both a handpump and an electric pump and that the remain- ing 38% have only a handpump, households will have already invested over Rs. 1 million in private handpumps and electric pumps. The operation and maintenance costs of these privately provided water systems is estimated to be

about Rs. 9800/month. The total monthly costs are about Rs. 23,900 (about Rs. 5/capita).

Data from the PHED in the Punjab indicate that a new piped distribution system for such a village would cost about Rs. 1.5 million (Rs. 300/capita). Based on these cost esti- mates, if 100% of the households in the village are connected and each pays Rs. 500 to connect their house to the distri- bution system, the total capital costs of the system would be approximately Rs. !.78 million. According to PHED esti- mates, the monthly operation and maintenance costs of such a piped distribution system would be about Rs. 3,800. The total monthly costs would be Rs. 21,600.

Given the approximate nature of these estimates, the total costs of a piped distribution system are essentially the same as the amount households are already spending for hand- pumps and electric pumps. There are numerous qualifica- tions to such a comparison. Among the most important of these arises from the fact that the estimated costs of the piped distribution system are based on the historical expe- hence of the PHED in building and operating water systems with low reliability. Such systems would not replace house- holds' need for private handpumps. Indeed, for some house- holds such systems might not even reduce their desire for electric pumps. Increasing the reliability of such systems would increase their costs. Hence a piped distribution sys- tem that would actually serve as a substitute for handpumps and electric pumps would likely cost more than households are now spending.

1950 ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN

• O0 2500

80 2000

g 60 15O0 '• 40 1000

• 20 t 500 0 0

0 10 20 30 40 50 60 70 80 90 100

Monthly Tariff (Rs. per month)

Brackish Water Zone

Villages without Piped Water Supply Standard Reliability

Fig. 1. Connection frequency and revenues versus monthly tariff.

Revenues

On the other hand, our estimates of what households are now spending miss some components, such as the cost of time household members spend collecting drinking and cooking water outside the home (see Whittington et al. [1990] for discussion of these issues in another developing country) or money paid to water vendors. It is hard to evaluate whether these additions would make the two alter-

native aggregate costs comparable. Also, the cost estimates for the piped water system assume that 100% of the house- holds would connect, and thus that the number of house- holds with service better than a handpump would be greater than in the current situation.

On average, households in villages in the brackish water zone without piped water supplies said they would be willing to pay Rs. 40/month for a connection to a standard piped water system (Table 4). If every household connected and paid Rs. 40/month, the monthly revenues to the water utility would be about Rs. 22,500, almost the same as our estimates of the total monthly costs of a piped distribution system and of current water expenditures. Moreover, our estimate of the total monthly costs included the costs of connecting the household to the distribution line. The water utility would not need revenues to cover these costs. However, Rs. 40/month is the mean WTP bid; not every household would be willing to pay this amount. Figure 1 indicates that if the monthly tariff were set at Rs. 25 (compared to the existing tariff of Rs. 12/month in the brackish water zone), about 80% of the households would decide to connect to a new piped distribution system. The total monthly revenue from a tariff of Rs. 25 would be about Rs. 11,700 (Figure 1 and Table 7). This would be more than three times the monthly operation and maintenance costs of the system to the PHED, and about two-thirds of the total monthly costs (including the capital costs).

If system reliability is improved, tariff levels can be raised even more and revenues can be significantly increased. An analysis of households' WTP bids suggests that 80% of the households indicated that they would be willing to pay Rs.

35/month for a connection to an improved, more reliable system. This would generate monthly revenues of more than Rs. 16,000 (Figure 2 and Table 7). Even allowing for the approximate nature of these estimates, such revenues are close to the total monthly costs of the piped distribution system. These calculations suggest that tariff levels in the brackish water zone could easily be doubled if reliability were improved. Moreover, the revenues generated would appear to be sufficient to pay for the operation and mainte- nance costs of the system as well as to make a contribution toward its capital costs.

These calculations are only illustrative. Our willingness- to-pay models in Table 5 indicate that there are differences in what specific households state that they would pay for connections to a reliable public system. Moreover, the costs

TABLE 7. Comparison of Costs, Revenues, and WTP Bids

Typical Village With 5,000 Population

(Brackish Water Zone)

Total monthly cost to PHED of piped water system if 100% of households connected 80% of households connected

Current household monthly water expenditures

Summation of households' WTP bids for standard system

Summation of households' WTP bids for improved system

Estimated revenue from a standard systema Estimated revenue from an improved system b

Rs. 18,800 Rs. 15,000 Rs. 23,900

Rs. 22500

Rs. 31,500

Rs. 11,700 Rs. 15,000

Estimated revenue from a standard system based on a tariff of Rs. 25/month, 80% of households connected, and Rs. 80 connection fee. Estimated revenue from an improved system based on a tariff of Rs. 35/month, 80% of households connected, and Rs. 80 connection fee.

ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN I951

2OOO

15O0

1000

5OO

0

0 10 20 30 40 50 60 70 80 90 100

Monthly Tariff (Rs. per month)

Brackish Water Zone

Villages with Piped Water System

Fig. 2. Revenues versus monthly tariff.

of such a system are likely to be higher than those currently available. Nonetheless, they do suggest that increasing reli- ability may be "affordable" because it enables the water authority to charge substantially higher tariffs and still main- rain high connection rates.

In the sweet water zone the costs of a water supply system are essentially the same as in the brackish water zone, but households are on average only willing to pay about half as much. Full cost recovery is thus not feasible in most villages in the sweet water zone at this time. However, households in the sweet water zone also place a premium on obtaining reliable service, and revenues can be increased by increasing reliability.

PROBLEMS WITH EXISTING GOVERNMENT POLICY

Many households in these two areas appear to be meeting their water needs through private-sector initiatives without reliance on government-subsidized projects. The govern- ment programs are generally regarded as unsuccessful at providing reliable water. We might ask whether these public sector programs have impacts beyond these failures. We believe that there are at least four problems with existing government policies in the rural water sector in the irrigated areas of the Punjab.

A Failure to Recognize the Common Property Nature of Groundwater

Existing government policy does not adequately address the consequences of the unregulated use of groundwater by households and small firms acting independently of each other. In many parts of the Punjab a serious problem is rising water tables that result from leaky irrigation canals, leading to waterlogging and salinity. At the same time in other locations the current practice of people drilling their own wells and pumping groundwater can cause falling water tables and localized overexploitation of the groundwater aquifer. Many households already have to rebore their wells regularly in order to reach water.

Not only does reboring add considerably to the costs of the private-sector option, but in many villages the ground- water is rapidly falling to a level at which it will become technically impossible to lift water with the type of simple, inexpensive handpump now prevalent in the Punjab. When this point is reached, villages will face a new kind of water problem. Poor households may be unable to afford the expense of both reboring and the installation of either an electric pump or a more costly, heavy-duty handpump required for the increased lifts. Such households may well have to purchase water from wealthier neighbors who can afford such systems. As the water table continues to fall, the economics of village water supply systems will shift dramat- ically in favor of collective solutions. Public water supply systems might still be based on groundwater but this could require large tubewells farther away from the village or town.

Deterioration of Sanitation and Drainage Situation

Second, existing government rural water policy does not pay adequate attention to sanitation and drainage issues. The increased water usage associated with the installation of electric pumps and indoor plumbing facilities has increased drainage problems in most villages. The open drains that line both sides of most streets carry sullage water and raw sewage. Often they are stagnant and overflow, creating public health problems.

Most village drainage systems empty into a sewage pond at the edge of the village. These ponds usually have a surface area of about half a hectare. They are typically unmanaged and are covered with vegetation. Wastewater from these sewage ponds evaporates, seeps into the groundwater, and sometimes drains into irrigation ditches. These ponds are rapidly filling up with solids as flows increase due to in- creased water use, installation of flush toilets, and popula- tion growth.

In both the sweet water and brackish water zones, the drainage situation will become increasingly worse with con-

1952 ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNSAB, PAKISTAN

tinued installation of electric pumps and flush toilets with holding tanks. Private-sector solutions to these kinds of drainage problems are not easy to envisage. Collective action at the community level is necessary to install drainage systems where they do not exist, to keep existing drains clean, and to find solutions to the problems associated with diminishing capacity of sewage ponds. This does not neces- sarily mean that large subsidies need to be provided by the central government for construction projects, but the public sector certainly has a role to play in providing technical assistance to communities in the areas of sanitation and

drainage planning.

Inequitable Distribution of Government Subsidies

Third, current policy results in large subsidies to villages in the brackish water zone that receive government-provided piped water systems. These subsidies are justified on the grounds that villages in the brackish zone are in greatest need of improved water. It is certainly true that these villages need improved quality water more than villages in the sweet water zone. However, this does not imply that because a village needs improved water, it must be poor. Many of the villages in the brackish zone that receive government subsidies are relatively high-income communi- ties, particularly those in the Faisalabad district that are engaged in textile enterprises. In fact, in our study, house- holds in villages in the brackish water zone were significantly better off than those in the sweet water zone.

Households in the brackish water zone have demonstrated

a high willingness to pay for improved water systems, both in terms of their current expenditures on electric pumps and in their responses to direct WTP questions. Because many villages in the brackish water zone appear to be able to afford to pay a substantial portion of the capital costs of reliable piped water systems, there seems to be no reason on equity grounds why systems in such relatively high-income com- munities should be heavily subsidized especially given the absence of progressive taxation in these rural areas.

Lack of Support for Economic Development

Fourth, government policy is not adequately supporting economic development in the irrigated areas of the Punjab. The results of this study show that in some large villages people are prepared to pay a much larger portion of costs of piped water supply systems if the service is reliable, even though they already have private handpumps. In the brack- ish water zone, although the official government policy is to promote the provision of piped water systems, in practice, few villages receive such systems because of the limited capital available from the central government to subsidize their construction. It makes no sense to delay providing service to communities in the brackish zone which do not require subsidies just because the central government does not have sufficient resources to subsidize 100% of the capital costs. Moreover, if some communities in the sweet water zone do not require subsidies, it makes no sense to deny them service just because their basic needs are being met by private handpumps.

One especially short-sighted aspect of current government policy is that it effectively denies the services of a piped water system to households and small firms in one of the

most dynamic economic regions of Pakistan. Water services are an important factor input for many small-scale manufa½. turing and commercial enterprises. Providing reliable piped water supplies to communities in the irrigated areas of the Punjab which are prepared to pay the costs of such services is a sound economic investment and will further spur the rapid ongoing development of such communities.

POLICY RECOMMENDATIONS

The findings from this research suggest several specific ways in which Pakistan's rural water policies can be ira. proved. People are prepared to pay substantial amounts for a private house connection, but only if the service is reliable. Households in the irrigated areas of the Punjab have several options with respect to what source of water they use. Unless a piped distribution system provides good service, many households will not connect to the system.

A manager of a public water authority must therefore compete for households' patronage. The most effective way to compete is not through subsidized tadifs, reduced con- nection fees, or low-cost financing of connection costs, but rather by providing a service that is better than households' other alternatives. Our findings indicate that the service characteristics of most concern to households are reliability and water quality.

Our analysis shows that households are willing to pay as much as 40% more per month for reliable service than for the existing (unreliable) service. Equally important, more house- holds will connect to a reliable system than to an unreliable system. The combination of both higher connection rates and the higher tadifs associated with improved reliability means that revenues to the utility can be increased signifi- cantly if reliability is improved (see Figure 2). Cost recovery and customer satisfaction thus both depend upon increasing the reliability of piped water systems. How can this be done?

The government-financed water systems built in the Pun- jab have been designed to accommodate an estimated per capita water use of 40 L/day. This estimate is much too low when the marginal cost of water to the household is zero, in part because humans are not the only users of the piped water. Many households in the villages own animals that live in their dwellings or compounds. (In the brackish zone, 40-50% of the sample households in a given village reported owning animals.) Water buffalos in particular need large amounts of water to stay cool. Traditionally, water buffalos were taken to a pond outside the village to cool off and drink, but this is a time-consuming activity. When a piped system is installed in a village, most people find it much more conve- nient simply to hose down their water buffalos in the street in front of their house. Because the connections are unme- tered, households have no incentive to conserve water.

In addition to watering and washing animals, households find many other uses for water when it is available essen- tially free (i.e., at zero marginal cost assuming the monthly tariff is paid). This high demand for water in a hot, arid climate when the marginal cost to the household is zero, coupled with the low capacity of the systems, means that water must be rationed by curtailing service. One result is chronically low pressure.

The Public Health Engineering Department and the com- munity are thus caught in a kind of low-level equilibrium trap [Singh et al., this issue]. Because water connections are not

ALTAF ET AL.: RURAL WATER SUPPLY POLICY iN PUNJAB, PAKISTAN 1953

metered, people demand more water than the system can provide. In order to ration supplies, the PHED reduces the number of hours of service. Because the systems are unre- liable, people must still invest in alternative private arrange- ments; the piped system becomes at best a supplemental water source. This means that households are not willing to pay very much for the service. Because households are not willing to pay much for the service, the PHED does not collect sufficient revenues to manage the system properly, and reliability deteriorates further.

If households can afford the higher costs associated with metered private connections, the way out of this trap is to install metered connections and charge higher prices for water. The higher prices will enable the PHED to collect the higher revenues required to pay the increased capital costs and provide more reliable service. The results of our survey suggest that this would be a popular policy change. The majority of respondents in both the sweet water and brackish water zones favored metering of private water connections. At first glance, this might seem unreasonable because house- holds now pay a subsidized tariff. However, although the tariff is low, so is the level of service. Households are expected to pay the tariff even when the system is down and they have not received any water. Households perceive metering to be a much more equitable arrangement: you pay for what you get, and you do not pay for what you do not get!

Of course, the decision on whether to install meters must take into careful consideration the cost of purchasing, main- taining, and reading the meters. However, in the past calculations of the costs and benefits of metering have often been too simplistic and failed to take account of the man- agement and strategic implications of a metering program. The absence of meters in an arid environment such as the

Punjab has profound negative consequences for the ability of a water authority to deliver reliable services in a riscally responsible manner. If the majority of households in a community cannot afford to pay for metered private connec- tions, the policy response should not be to simply provide unmetered connections. Rather, the entire package of water services needs to be reexamined [World Bank Research Team, 1993].

To summarize, a sound rural water policy for the irrigated areas of the Punjab should be based on the following principles.

1. Improved reliability is the key to customer satisfac- tion and high connection rates. The provision of subsidized credit to cover household connection costs is not necessary.

2. Private connections should be metered, and prices should be set to cover a much higher proportion of the capital and operation costs of supplying water. Uniform prices are not desirable. They should vary based on system costs.

3. Piped distribution systems should not be built unless projected revenues are sufficient to cover the costs of providing much more reliable service.

4. Systems should not be designed on the basis of a 40 L per capita per day standard. Information is required on how much water households will consume with metered connec-

tions and realistic prices. 5. Villages with populations of less than 5000 should not

be prohibited from obtaining piped water distribution sys- tems, provided that they can pay for the costs of supply.

6. Water systems based on distribution by public taps

can rarely be economically justified in the irrigated areas of the Punjab. They should only be constructed when there is clear, demonstrable evidence that this is the service level

desired by the community and that people are prepared to pay for the costs of the system.

7. Public sector efforts should be focused on communi-

ties where the problems of declining groundwater levels and deteriorating groundwater quality are most imminent.

8. Villages in the sweet water zone should not be pre- cluded from obtaining piped water distribution systems, provided that they are able and willing to pay for the costs of supply.

9. Adequate drainage should be included in the design of public water supply systems.

The actual implementation of these principles will require a redefinition of the proper roles of government and the community in the water sector in Pakistan. A detailed discussion of this topic is outside the scope of this paper. It is clear, however, that the institutional capacity does not currently exist on the part of government to undertake the kind of technical assistance required to address the needs and conditions of villages on a community-by-community basis (see A. Hasan and Y. Sheikh, unpublished manuscript, !985, and Hasan [1986] for examples). Efforts would be required to decentralize by making communities stronger political entities, to provide better central government tech- nical support, to develop groundwater pumping laws or other allocative mechanisms, and/or to develop a regulatory structure for water providers as public utilities.

CONCLUDING REMARKS

The existing government policies for the rural water sector in the irrigated areas of the Punjab need to be reevaluated in light of the findings from this study. Public-sector agencies should not assume that households in the large villages in this area that have not yet received piped water systems have been waiting patiently for a government rural water supply project to be initiated. On the contrary, because the demand for improved services is high, the private sector has provided many people with alternative solutions to their water needs. This means that the PHED no longer has carte blanche to come into a village and decide what service level and system reliability characteristics to provide. For exam- ple, there is little demand for public standposts in the study villages and a strong desire for metered connections with reliable pressure.

Planning for improvements in rural water supplies now requires a detailed knowledge of household demand for improved services and active community involvement in the decision-making process. Otherwise, services may be pro- vided that households do not want, will not use efficiently, and will not pay for. Perhaps most importantly, the results of this study suggest that circumstances in large villages in the irrigated areas of the Punjab are changing rapidly, and government planning efforts must be much better informed about the nature of these changes and their implications for policy.

APPENDIX

This research was part of a multicountry study funded by the World Bank. The overall research design is described in

1954 ALTAF ET AL.: RURAL WATER SUPPLY POLICY IN PUNJAB, PAKISTAN

the paper summarizing the results from all the field studies [Wor M Bank Research Team, 1993].

The research design called for surveys in three different econoenvironmental zones. Within each zone representative villages of two types were to be selected: type A, with piped water systems, and type B, without piped water systems but where installation was scheduled in the near future. Approx- imately 200 households were to be randomly sampled within each village type.

In the Punjab, the type B villages were further discrimi- nated into the following two subtypes: type B1, currently without piped systems but where installation was scheduled in the near future, and type B2, currently without piped systems and with no prospects of installation in the near future. This was to test for strategic bias in the willingness- to-pay bids of the respondents.

A bidding game format (for details see Mitchell and Carson [ 1989, pp. 99-100] was used to elicit the willingness- to-pay bids. An extended bidding game structure (allowing for more than two iterations) was used as this was compat- ible with the practice of bargaining in the Punjab. High and low starting point versions of the bidding games were used to test for starting point bias. The starting points were Rs. 15 and Rs. 50 in the sweet water zone and Rs. 20 and Rs. 40 in the brackish water zone.

Acknowledgments. We thank the following individuals for their comments and suggestions on a previous draft of this paper: Arif Hasan, Donald T. Lauria, John Briscoe, Rick Pollard, Roshaneh Zafar, and Azmat Isa. Funding for this research was provided by the World Bank and the United Nations Development Program's Water and Sanitation Project.

REFERENCES

Alta/, M. A., H. Jamal, J. L. Liu, V. K. Smith, and D. Whittington, The willingness to pay for rural water connections in developing countries: A discrete choice analysis for Pakistan, PRE draft report, 44 pp., Infrastructure and Urban Dev. Dep., World Bank, Washington, D.C., 1992a.

Altaif, M. A., H. Jamal, and D. Whittington, Willingness to pay for

water in rural Punjab, Water Sanit. Rep. 4, 161 pp., UNDP-World Bank Water and Sanit. Program, Washington, D.C., 1992b.

Fass, S., Political Economy in Haiti: The Drama of Survival. Transaction Books, New Brunswick, N.J., 1988.

Hasan, A., Field trip report for Northwest Frontier Province: Findings and reflections, UNICEF water assessment for Pakistan, UNICEF Mission, Karachi, Pakistan, 1986.

Mitchell, R., and R. Carson, Using Surveys to Value Public Goods: The Contingent Valuation Method, Resources for the Future, Washington, D.C., 1989.

Pasha, H., and M. McGarry, Rural water supply and sanitation in Pakistan: Lessons from experience, World Bank Tech. Pap. 105, World Bank, Washington, D.C., 1989.

Singh, B., R. Ramasubban, R. Bhatia, J. Briscoe, C. Griffin, and C. Kim, Rural water supply in Kerala, India: How to emerge from a low-level equilibrium trap, Water Resour. Res., this issue.

Smith, V. K., J. L. Liu, M. A. Altaf, H. Jamal, and D. Whittington, How reliable are contingent valuation surveys for policies in developing economies, draft report, Dep. of Econ., N. C. State Univ., Raleigh, 1991.

Stewart, M. B., On least squares estimation when the dependent variables are grouped, Rev. Econ. Stud., 50, 737-754, 1983.

Whittington, D., D. T. Lauria, D. A. Okun, and X. Mu, Water vending in developing countries, Int. J. Water Resour. Dev., 5{31, 158-168, 1989.

Whittington, D., X. Mu, and R. Roche, Calculating the value of time spent collecting water: Some estimates for Ukunda, Kenya, World Dev., 18(2), 269-280, 1990.

Whittington, D., D. T. Lauria, and X. Mu, A study of water vending and willingness to pay for water in Onitsha, Nigeria, World Der., 19(2/3), 179-198, 1991.

World Bank Research Team, Demand for water in rural areas: Determinants and policy implications, World Bank Res. Observ., 8(1), 47-70, 1993.

M. A. Altaf and H. Jamal, Applied Economics Research Centre, University of Karachi, P.O. Box 8403, Karachi-75270, Pakistan.

V. K. Smith, Resource and Environmental Economics Program, North Carolina State University, Raleigh, NC 27599.

D. Whittington, Departments of Environmental Sciences and Engineering, and City and Regional Planning, CB 3140, New East Building, University of North Carolina at Chapel Hill, Chapel Hi!I, NC 27599.

(Received February 8, 1991; revised March 18, 1992; accepted April 27, 1992.)